Long-Term Cryopreservation Preserves Blood–Brain Barrier Phenotype of iPSC-Derived Brain Microvascular Endothelial Cells and Three-Dimensional Microvessels

Abstract

Brain microvascular endothelial cells derived from induced pluripotent stem cells (dhBMECs) are a scalable and reproducible resource for studies of the human blood–brain barrier, including mechanisms and strategies for drug delivery. Confluent monolayers of dhBMECs recapitulate key in vivo functions including tight junctions to limit paracellular permeability and efflux and nutrient transport to regulate transcellular permeability. Techniques for cryopreservation of dhBMECs have been reported; however, functional validation studies after long-term cryopreservation have not been extensively performed. Here, we characterize dhBMECs after 1 year of cryopreservation using selective purification on extracellular matrix-treated surfaces and ROCK inhibition. One-year cryopreserved dhBMECs maintain functionality of tight junctions, efflux pumps, and nutrient transporters with stable protein localization and gene expression. Cryopreservation is associated with a decrease in the yield of adherent cells and unique responses to cell stress, resulting in altered paracellular permeability of Lucifer yellow. Additionally, cryopreserved dhBMECs reliably form functional three-dimensional microvessels independent of cryopreservation length, with permeabilities lower than non-cryopreserved two-dimensional models. Long-term cryopreservation of dhBMECs offers key advantages including increased scalability, reduced batch-to-batch effects, the ability to conduct well-controlled follow up studies, and support of multisite collaboration from the same cell stock, all while maintaining phenotype for screening pharmaceutical agents.

Graphical Abstract

INTRODUCTION

The human blood–brain barrier (BBB) controls cellular and molecular transport into and out of the brain.¹ The brain microvascular endothelial cells (BMECs) that form the cerebrovasculature express tight junctions, efflux pumps, and nutrient transporters, which regulate both paracellular and transcellular permeability. Recently, induced pluripotent stem cell (iPSC)-derived human BMECs (dhBMECs) have emerged as a scalable and reproducible source for in vitro models of the human BBB.^2–5 For example, dhBMECs are a resource for screening pharmaceuticals for brain penetration^6,7 and studying mechanisms of drug delivery.⁸ As dhBMECs cannot be passaged without phenotypic drift, cryopreservation can enable storage of cells for many applications.⁹ Additionally, stem cell differentiations are prone to batch-to-batch variability due to inconsistencies in reagents and iPSC seeding density. While many efforts are taken to reduce variability, including using defined reagents and seeding densities,^10–12 cell cryopreservation eliminates the need to perform unique differentiations for each experiment. Successful implementation of cryopreserved dhBMECs in two-dimensional (2D) and three-dimensional (3D) models will significantly improve the scalability of BBB research.

Here, we build on initial reports of dhBMEC cryopreservation to address the long-term changes in BBB phenotype after 1 year of cryopreservation. In a previous study, immunocytochemistry images showed no differences in expression and localization of key BBB markers in dhBMECs thawed in media containing Rho-associated protein kinase inhibitor Y27632 (ROCK inhibitor) after 2 days of cryopreservation.⁹ Furthermore, transendothelial electrical resistance (TEER) of confluent monolayers remained high following 1-year cryopreservation.⁹ Here, we report on gene and protein expression, TEER, and permeability of various solutes for dhBMEC monolayers cryopreserved for up to 1 year in a 2D model and in a 3D tissue engineered model of the BBB. In general, 2D monolayers of cryopreserved dhBMECs maintained their phenotype compared to freshly differentiated cells. There were no significant changes in expression and localization of tight junction proteins (claudin-5, occludin), nutrient transporters (GLUT-1), efflux pumps (P-gp), and endothelial markers (CD31, VE-cadherin) after 1-year cryopreservation. Using bulk RNA sequencing, we found remarkable similarity between fresh and 1-year cryopreserved cells, with no differentially expressed genes (DEGs). Similarly, the permeabilities of rhodamine 123, glucose, and 10 kDa dextran across 2D confluent monolayers were unchanged following 1-year cryopreservation. However, we found two differences associated with cryopreservation: (1) a gradual decrease in the yield of adhesive cells following thawing and (2) a larger decrease in TEER in confluent monolayers following a medium change in a transwell compared to non-cryopreserved cells, which also resulted in higher Lucifer yellow permeability. To assess functionality in tissue engineered models, cryopreserved dhBMECs were incorporated into a 3D model of the BBB.^5,13–15 Within the 3D model, cryopreserved dhBMECs maintained functionality and were not vulnerable to stress caused by a media switch as in 2D models. Long-term cryopreserved dhBMECs are a suitable cell source for studies of drug delivery in both 2D and 3D BBB models.

EXPERIMENTAL SECTION

Differentiation.

dhBMECs were generated from the BC1 iPSC line derived from a 46 year old male, as previously reported,^14,15 using sequential treatment with feeder-free iPSC maintenance media (TESR-E8; Stem Cell Technologies), unconditioned media without bFGF (UM/F-), and endothelial media (outlined below). The first 6 days of differentiation were conducted in UM/F-: DMEM/F12 (Life Technologies) supplemented with 20% knockout serum replacement (Life Technologies), 1% non-essential amino acids (Life Technologies), 0.5% GlutaMAX (Life Technologies), and 0.836 μM beta-mercaptoethanol (Life Technologies). The final 2 days of differentiation were conducted in endothelial media: human endothelial cell serum-free medium (HESFM) (ThermoFisher Scientific) supplemented with 1% human platelet poor derived serum (Sigma-Aldrich), 2 ng mL⁻¹ bFGF (Fisher Scientific), and 10 mM all-trans retinoic acid (Sigma-Aldrich). At the end of differentiation, cells were singularized using a 10 min treatment with accutase (Invitrogen) at 37 °C and then used fresh or after cryopreservation (Figure 1a). For fresh use, cells were seeded at 1 × 10⁶ cells cm⁻² on surfaces coated overnight with 50 μg mL⁻¹ human placental collagen IV (Sigma) and 25 μg mL⁻¹ fibronectin from human plasma (Sigma). For cryopreservation, dhBMECs were resuspended into 60% endothelial media, 30% fetal bovine serum (Invitrogen), and 10% dimethyl sulfoxide (MilliporeSigma). Vials were chilled to −80 °C using an isopropanol-filled Mr. Frosty container (Nalgene) for 24 h and then transferred into liquid nitrogen vapor for long-term storage. To thaw, vials were quickly warmed in a water bath at 37 °C, diluted 10-fold in endothelial media, and centrifuged to a pellet. To isolate viable endothelial cells, all cells were subcultured on a collagen IV and fibronectin-coated surface for 1 h using endothelial media supplemented with 1% penicillin–streptomycin (ThermoFisher Scientific) and 10 μM ROCK inhibitor Y27632 (ATCC). During this period, endothelial cells were adherent, while other cells produced by the differentiation along with non-viable cells did not adhere.⁴ Cell viability was determined via exclusion of Trypan blue (Corning). Following subculture, dhBMECs were detached by treating with accutase for 10 min and then seeded at 1 × 10⁶ cells cm⁻² on collagen IV and fibronectin-coated surfaces, matching densities used with fresh cells. For all assays, cells were cultured for the first 24 h after seeding in endothelial media supplemented with 1% penicillin–streptomycin (and 10 μM ROCK inhibitor only for cryopreserved cells), before switching to basal media: HESFM supplemented with 1% human platelet poor derived serum and 1% penicillin–streptomycin. For a limited number of experiments, dhBMECs were formed from the WTC iPSC line¹⁶ with green fluorescent protein-tagged zona occludens-1 or β-actin (Allen Cell Institute).

Figure 1.

Cryopreserved dhBMECs maintain high transendothelial electrical resistance (TEER) for over 1 year. (a) Differentiation and cryopreservation of iPSC-dhBMECs. (b) Subculture of non-frozen, 1-day cryopreserved, and 1-year cryopreserved dhBMECs. (c,d) Cryopreservation is associated with a loss of cell adherence and slight decrease in cell viability. Data collected across n = 3 independent differentiations for each condition. (e) TEER of cryopreserved dhBMECs versus the length of cryopreservation. n = 37 unique thaws across 22 independent dhBMEC differentiations. (f) Fractional TEER (%) displaying the ratio of cryopreserved to fresh TEER over cryopreservation length. Data collected across n = 19 unique thaws consisting of n = 10 unique dhBMEC differentiations. All data represent averages across 2–6 technical replicates (individual transwells).

TEER Measurement.

TEER is a measure of barrier integrity of cell monolayers, with theoretical and animal model values ranging from 1500 to 8000 Ω cm².⁵ TEER was recorded using an EVOM2 (World Precision Instruments), as previously reported.¹³ Transwells (6.5 mm) with a 0.4 μm pore polyester membrane insert (Corning) were used for all measurements. All recordings were conducted immediately after removal from an incubator to prevent fluctuations in monolayer resistance due to temperature changes. TEER values were corrected for the value of a transwell insert without cells and normalized by monolayer surface area.

Immunocytochemistry.

Immunocytochemistry was used to identify expression and localization of BBB and endothelial markers. Fresh and 1-year cryopreserved cells were cultured for 2 days on borosilicate cover glass slides, seeded at 250,000 cells cm⁻². Cells were washed with phosphate-buffered saline (PBS; ThermoFisher), fixed with methanol for 15 min, and then blocked with 10% goat serum (Cell Signaling Technology) or 10% donkey serum (Millipore Sigma), supplemented with 0.3% Triton X-100 (Millipore Sigma) in PBS for 30 min. Primary antibodies were used as previously reported.¹³ Alexa Fluor-647 and Alexa Fluor-488 secondary antibodies (Life Technologies) were diluted 1:200 in blocking buffer for 45 min at room temperature. To localize nuclei, cells were treated with 1 μg mL⁻¹ DAPI (ThermoFisher). Images were acquired at 40× magnification using a swept field confocal microscope (Prairie Technologies) and illumination supplied by an MLC 400 monolithic laser combiner (Keysight Technologies). To control for non-specific binding of secondary antibodies, images were collected in the absence of primary antibodies; all stains were above background levels, indicating presence of primary protein targets.

Permeability Assays.

The permeability of 200 μM Lucifer yellow (ThermoFisher), 2 μM Alexa Fluor 647-conjugated 10 kDa dextran (ThermoFisher), 10 μM rhodamine 123 (ThermoFisher), and 25 mM D-glucose (Sigma) across dhBMEC monolayers was measured 2 or 3 days after seeding on transwells. Transport assays were conducted in transport buffer comprised of distilled water with 0.12 M NaCl, 25 mM NaHCO₃, 2 mM CaCl₂, 3 mM KCl, 0.4 mM K₂HPO₄, 2 mM MgSO₄, and 1 mM HEPES. Transport buffer for Lucifer yellow, 10 kDa dextran, and rhodamine was supplemented with 0.1% human platelet poor derived serum. To measure apical-to-basolateral permeability, transport buffer with the compound of interest was added to the apical chamber (100 μL), while the basolateral chamber contained 600 μL of transport buffer. To measure the basolateral-to-apical permeability, transport buffer supplemented with the compound of interest was added to the basolateral chamber (600 μL), while the apical chamber contained 100 μL of transport buffer. For Lucifer yellow and 10 kDa dextran transport measurements, apical-to-basolateral permeability was measured at 15 and 30 min. For rhodamine 123, both apical-to-basolateral and basolateral-to-apical permeability were measured at 60 and 90 min. For glucose, apical-to-basolateral permeability was measured at 5, 10, 15, and 30 min, as previously reported.¹⁷ All time points were selected to ensure that the permeability remained within the linear regime.¹ Apparent permeability was calculated as P = (dC/dt) (V·1/A) (1/C₀), where dC/dt is the slope of cumulative concentration, V is the volume of the receiving compartment (i.e., basolateral or apical chamber), A is the area of the monolayer, and C₀ is the dosed concentration of solute.¹⁸ All transport assays were conducted under gentle rocking at 37 °C to ensure rapid equilibration of the compounds of interest. At each time point, media in apical and basolateral compartments were collected and stored in a 96-well plate for analysis using a Synergy H4 microplate reader (Biotek) for Lucifer yellow, 10 kDa dextran, and rhodamine. Glucose samples were collected and frozen at −20 °C prior to analysis, then thawed for 2 h at room temperature and transferred to 96-well plates. The following excitation and emission settings were used for analysis: Lucifer yellow (428 nm/545 nm), 10 kDa dextran (647 nm/667 nm), and rhodamine 123 (495 nm/525 nm). Glucose transport was quantified using a glucose colorimetric detection kit (ThermoFisher Scientific), utilizing absorbance measurements at 560 nm. Concentrations of all probes were determined from calibration curves based on serial dilution of each probe spanning 4 orders of magnitude. The apparent permeability of each probe was calculated as previously reported.^3,17 For 10 kDa dextran and rhodamine 123, efflux ratios were calculated as the ratio of basolateral-to-apical and apical-to-basolateral permeability. Biological replicates of permeability measurements were averaged across at least two technical replicates (i.e., individual transwells).

RNA Sequencing.

Three biological replicates of BC1 iPSC colonies, freshly differentiated BC1 dhBMECs, and 1-year cryopreserved BC1 dhBMECs were analyzed. BC1 iPSC colonies were harvested prior to the UM/F- stage of differentiation; BC1 dhBMECs were harvested as confluent monolayers following subculture on collagen IV and fibronectin-coated plates. Six-well plates were seeded at a 1:1 area ratio from the differentiation and cultured for 1 day in endothelial media supplemented with 1% penicillin–streptomycin and 10 mM ROCK inhibitor, and then 1 day in basal media (matching the day 2 time point in transwell assay). Cells were washed with 1× PBS and lysed using RLT buffer supplemented with β-mercaptoethanol, before isolating RNA using a RNeasy Mini Kit (Qiagen) with DNase I digestion (Qiagen). All RNA samples had an RNA integrity number >9.7 as measured by an Agilent 2100 bioanalyzer. Total RNA was subjected to oligo (dT) capture and enrichment, and the resulting mRNA fraction was used to construct cDNA libraries. Sequencing was carried out on an Illumina NovoSeq platform (performed by Novogene) with paired end 150 bp reads, generating approximately 20 million paired reads per sample. The R package Rsubread (Version 2.0.1) was used for raw read alignment and for read quantification to the reference human genome (GRCh38).¹⁹ Filtering of raw read counts was performed using the filterByExpr function in R package edgeR (Version 3.28.1).^20,21 Pearson correlation coefficients were computed based on filtered CPM gene expression. Differentially expressed genes (DEGs) were determined using the R package limma (Version 3.42.2).²² The normalization factor for each sample was first calculated using the calcNormFactors function from edgeR. The voom function from limma was then used in conjunction with these factors, as well as the filtered CPM counts, to test for differential expression, where an adjusted p-value (false discovery rate threshold calculated by the Benjamini–Hochberg method) of 0.05 was considered statistically significant. Gene set enrichment analysis was conducted on DEGs using DAVID (v.6.8).²³ Data are deposited at NCBI under the accession number GSE151976.

3D Microvessels.

Tissue engineered microvessels were fabricated using dhBMECs as previously reported.¹⁵ Briefly, 150 μm channels were patterned in 7 mg mL⁻¹ type I collagen hydrogels housed within PDMS. Channels were treated with 20 mM genipin (Wako Biosciences) for 2 h, washed in 1× PBS overnight, and then coated with collagen IV and fibronectin. dhBMECs were seeded at 1 × 10⁶ cells mL⁻¹ into channels and allowed to adhere for 20 min. Channels were then perfused at ~1 dyne cm⁻² using a gravity-driven syringe setup.¹³ 10× magnification epifluorescence and phase contrast images were collected using an inverted microscope (Nikon Eclipse Ti-E) maintained at 37 °C. After 2 or 3 days, microvessels were perfused with 200 μM Lucifer yellow or 2 μM 10 kDa dextran for 1 h to assess permeability.¹⁵ Intensity profiles of a region consisting of the microvessel lumen and surrounding extracellular matrix were obtained using ImageJ (NIH). Permeability was calculated as P = (d/4) (1/ΔI) (dI/dt), where d is the vessel diameter, ΔI is the initial increase in fluorescence intensity upon luminal filling, and (dI/dt) is the rate of increase in fluorescence intensity as the solute exits into the gel.^24,25

Statistical Analysis.

All statistical analyses were performed using Prism ver. 8 (GraphPad). Metrics are presented as mean ± standard deviation. A student’s unpaired t-test (two-tailed with unequal variance) was used for comparison of two conditions, and an analysis of variance (ANOVA) for comparison of three or more conditions. For ANOVA tests, reported p values were multiplicity adjusted using a Tukey test. Linear regression was conducted using least squares fitting with no constraints, with an F test used to determine if fitted slopes were statistically significant. Differences were considered statistically significant for p < 0.05, with the following thresholds: *p < 0.05, **p < 0.01, ***p < 0.001.

RESULTS

Time Course of Barrier Function following Cryopreservation.

Cells were cryopreserved using a previously reported strategy⁹ (Figure 1a). The differentiation of dhBMECs from iPSCs was carried out over 11 days as previously reported by multiple independent laboratories.^2,3,6,7 As a control for cryopreservation, freshly differentiated cells were immediately purified via subculture onto collagen IV and fibronectin-coated surfaces including transwell inserts for measurement of permeability and TEER, glass for immunocytochemistry, plastic 6-well plates for RNA sequencing, and collagen I microchannels for 3D microvessel studies. Differentiated cells were also cryopreserved and stored in liquid nitrogen for durations from 1 day to 1 year.

Cryopreserved cells were rapidly thawed and subcultured onto collagen IV and fibronectin-coated 6-well plates to purify viable dhBMECs. The yield of dhBMECs following the purification step was found to be dependent on the duration of cryopreservation. For comparison, non-frozen cells formed nearly confluent monolayers during subculture, indicating a high purity of endothelial-like cells, as validated in a previous work.^2,3 However, the population of adherent cells decreased after just 1 day of cryopreservation, and further decreased at 1 year (Figure 1b). The percentage of adherent cells on surfaces coated with basement membrane proteins after 1 h of subculture decreased from 90 ± 3% (no cryopreservation) to 76 ± 5% after 1 day of cryopreservation (p = 0.013) (Figure 1c). The percentage further decreased to 53 ± 4% after 1 year of cryopreservation (p = 0.001). The fraction of non-viable cells in the non-adherent population was determined from absorption of Trypan Blue. Although the fraction was very low across all conditions (<4%), it increased following 1-year cryopreservation compared to that in fresh (p = 0.008) but was not significantly different between 1 day and 1 year (p = 0.247) (Figure 1d). These results suggest that loss of cell adhesion was the primary mode of changes in dhBMEC purity, not loss of viability.

Because of the decrease in yield of adherent cells with duration of cryopreservation and variability in attachment between replicates, the seeding density was adjusted to achieve a consistent density of adherent cells. At constant cell density, TEER slightly decreased from 1934 ± 711 Ω cm² for fresh dhBMECs to 1425 ± 452 Ω cm² after 1-year cryopreservation (Figure 1e). A progressive decrease in TEER was confirmed from a statistically non-zero slope (p = 0.043, r² = 0.10). The fractional change in TEER of cryopreserved dhBMECs was obtained by normalizing to the value for non-frozen cells from the same differentiation. This linear relationship was not statistically significant (p = 0.086, r² = 0.16), although the high variability suggests that distinct differentiations were susceptible to variation in cryopreservation efficiency (Figure 1f).

BBB Phenotype after 1 Year of Cryopreservation.

Since previous studies focused on phenotypic characterization for short cryopreservation periods, we sought to assess BBB phenotype after 1 year. Non-frozen and 1-year cryopreserved dhBMECs were compared using various functional assays. Importantly, dhBMECs maintained expression and localization of tight junctions (claudin-5, occludin), nutrient transporters (GLUT-1), efflux pumps (Pgp), and endothelial markers (CD31, VE-cadherin) after 1 year of cryopreservation (Figure 2a). Claudin-5 and occludin displayed junctional localization, GLUT-1 and P-gp displayed membrane localization, and endothelial markers displayed junctional and some cytoplasmic localization. Although there was no difference in expression and localization of tight junction proteins, peak TEER decreased ~25% (p = 0.026) (Figure 2b). Generally, cryopreserved cells displayed peak TEER 72 h after subculture compared to fresh cells which reliably peak after 48 h (data not shown); this delay was independent of seeding density, as confirmed from analysis of nucleus density (Figure 2a).

Figure 2.

Phenotypic characterization of dhBMECs following 1-year cryopreservation. (a) Immunocytochemistry of tight junctions (claudin-5, occludin), nutrient transporters (GLUT-1), efflux pumps (P-gp), and endothelial markers (CD31, VE-cadherin). Representative images at day 2 after seeding are shown across n = 2 independent differentiations for each condition. (b) TEER values for non-frozen (n = 17) and 1-year cryopreserved (n = 7) dhBMECs. n represents the number of unique differentiations. (c) Permeability comparison of four compounds: Lucifer yellow (444 Da), 10 kDa dextran, glucose (180 Da), and rhodamine 123 (380 Da). Data represent independent differentiations for each condition, with n = 7–8 for Lucifer yellow and 10 kDa dextran, and n = 3–5 for rhodamine 123 and glucose. For independent differentiations, data represent averages across 2–6 technical replicates (individual transwells). *p < 0.05.

To probe the function of tight junctions, efflux pumps, and nutrient transporters, we compared the permeability of multiple compounds between fresh and 1-year cryopreserved dhBMECs. Following 1-year cryopreservation, the apical-to-basolateral permeabilities of glucose, 10 kDa dextran, and rhodamine were not significantly different from values obtained from non-frozen cells (p > 0.05 for all comparisons) (Figure 2c). However, the apical-to-basolateral permeability of Lucifer yellow, a paracellular marker, increased ~3-fold, from 2.18 ± 0.90 × 10⁻⁶ cm s⁻¹ (non-frozen) to 5.97 ± 2.49 × 10⁻⁶ cm s⁻¹ after 1-year cryopreservation (p = 0.001).

Additionally, we found that cryopreserved dhBMECs maintained functionality of two important transcellular transport pathways: p-glycoprotein (P-gp)-mediated efflux and glucose transporter-1 (GLUT-1)-mediated nutrient transport. The efflux ratio is a measure of directional transport due to polarization of transporters at the luminal and/or abluminal membrane.⁵ Rhodamine 123 is a lipophilic P-gp substrate, while 10 kDa dextran is a negative control due to its large size and lack of transporter specificity.²⁶ Rhodamine 123 permeability was higher in the basolateral-to-apical direction (p < 0.001 for non-frozen and frozen), while 10 kDa dextran permeability was similar in both directions indicating no directional transport (p > 0.05 for non-frozen and frozen) (Figure 2c). Efflux ratios for rhodamine 123 were 7.18 ± 1.71 for non-frozen cells and 9.17 ± 2.56 after 1-year cryopreservation (p = 0.268), indicating that cells maintain directional transport of P-gp substrates. Glucose permeability into the brain is regulated primarily by the expression of GLUT-1. The permeability of glucose was approximately 1 × 10⁻⁵ cm s⁻¹ for both non-frozen and 1-year cryopreserved cells (p = 0.824) (Figure 2c), indicating maintenance of GLUT-1 function.

Mechanisms of Changes in Barrier Function.

While the permeabilities of rhodamine, glucose, and 10 kDa dextran were unchanged after 1-year cryopreservation, there was a small increase in Lucifer yellow permeability, and the peak TEER decreased monotonically with the duration of cryopreservation. We hypothesized that the increase in Lucifer yellow permeability was related to stress induced by replacing medium in the apical chamber with transport buffer containing the compound of interest (Lucifer yellow, dextran, etc.) at the beginning of transwell permeability assays.

We first tested whether media switches were associated with alterations in expression or localization of zona occludens-1 (ZO-1) and β-actin in dhBMECs derived from the WTC line following cryopreservation. However, tight junction localization and actin cytoskeleton organization remained unchanged following a media switch and remained stable over an additional 30 min in transport buffer (Figure 3a).

Figure 3.

Response of cryopreserved dhBMECs to a media switch. (a) Cryopreserved dhBMECs generated from the WTC iPSC line do not display changes in actin or tight junction localization due to a media switch. Representative images are shown across n = 2 independent differentiations for each condition. (b) Percentage drop of TEER immediately following a media change to transport buffer. Data collected across n = 6 independent differentiations for each condition. (c) Time course of TEER following a media switch. Data collected across n = 3 independent differentiations for each condition. Permeability (apical-to-basolateral) for Lucifer yellow plotted versus (d) TEER recorded before the permeability measurement (no media switch) and (e) TEER recorded at the beginning of the permeability assay immediately following the media switch. Data collected across n = 21 individual transwells representing n > 5 independent differentiations of fresh or 1-year cryopreserved dhBMECs. *p < 0.05.

Interestingly, previous studies have suggested that the permeabilities of fluorescein (332 Da) and IgG (150 kDa) across dhBMEC monolayers are independent of TEER for values greater than about 900 and 500 Ω cm², respectively.⁶ Here, we observed a statistically significant increase in the permeability of Lucifer yellow across cryopreserved cells compared to non-frozen controls despite TEER values nearly double these thresholds. Therefore, we next measured TEER before and immediately after switching media to transport buffer prior to the start of each transwell permeability assay. For non-frozen dhBMECs, TEER values decreased to ~21% of the peak value following the media change, whereas 1-year cryopreserved cells decreased to ~10% of the peak value (p = 0.048) (Figure 3b). Additionally, throughout the 30 min time course of permeability measurements, cryopreserved cells maintained lower TEER values (~200 Ω cm²) than fresh dhBMECs (~400 Ω cm²) (Figure 3c). Thus, cryopreserved dhBMECs were more stressed after switching to transport buffer. This stress may contribute to differences in the permeability of Lucifer yellow, although not affecting the permeability of rhodamine, glucose, or 10 kDa dextran due to the low molecular weight and paracellular pathway of this dye.

In further support of this hypothesis, we assessed the correlation between TEER values recorded before and after the start of permeability measurements (before and after the media switch) and the subsequent permeability values. The correlation between TEER recorded immediately before a measurement and the subsequent Lucifer yellow permeability value showed a low coefficient of determination for both linear (r² = 0.13) and one-phase decay fits (r² = 0.14) (Figure 3d). These results indicate that there was limited correlation between the initial TEER and Lucifer yellow permeability. Similar results were obtained for 10 kDa dextran (results not shown). The correlation between TEER recorded at the beginning of a permeability measurement (after the media switch) and Lucifer yellow permeability displayed a high coefficient of determination for both linear (r² = 0.44) and one-phase decay fits (r² = 0.84). One-phase decay fits match the previous reports using dhBMECs.⁶ Similar results were obtained for 10 kDa dextran (r² = 0.34). These results suggest that the TEER value following a media switch is a strong predictor of permeability and that a unique drop in TEER is responsible for elevating Lucifer yellow permeability of cryopreserved dhBMECs.

Changes in Gene Expression following 1-Year Cryopreservation.

Bulk RNA sequencing was utilized to compare global gene expression profiles of source iPSCs to fresh and 1-year cryopreserved dhBMECs (Figure 4). Overall, gene expression of BC1 iPSCs displayed low Pearson correlation coefficients to fresh (ρ = 0.73) and cryopreserved (ρ = 0.74) dhBMECs, indicating distinct gene expression profiles. In contrast, the Pearson correlation coefficient between fresh and cryopreserved dhBMECs (ρ = 0.98) indicates a strong association between these two conditions. Additionally, Euclidian hierarchical mapping identified no clustering of cryopreserved and dhBMEC biological replicates, indicating that cryopreservation does not result in a distinct profile of gene expression compared to the normal variability between differentiation batches (Figure 4a). Gene expression for source iPSCs from across independent differentiations clustered closely and displayed numerous differentially expressed genes (DEGs) compared to dhBMECs (Figure 4b). Genes upregulated during differentiation include BBB markers (SLC2A1), endothelial markers (CDH5), and those previously identified as critical to barrier tightness (RARA),²⁷ while downregulated genes included transcription factors associated with pluripotency (POU5F1, SOX2, MYC). DEGs upregulated in dhBMECs highlight enrichment of processes and pathways associated with BBB function, including cell adhesion (GO:0007155; p = 1.13 × 10⁻⁷), Wnt signaling pathway (GO:0016055; p = 3.60 × 10⁻⁷), extracellular matrix organization (GO:0030198; p = 1.39 × 10⁻⁶), angiogenesis (GO:0001525; p = 1.11 × 10⁻³), cellular response to retinoic acid (GO:0071300; p = 1.22 × 10⁻³), and blood vessel development (GO:0001568; p = 2.72 × 10⁻³). No DEGs were identified between fresh and cryopreserved dhBMECs (Figure 4c), indicating highly similar gene expression profiles.

Figure 4.

Gene expression is maintained following 1-year cryopreservation of dhBMECs. (a) Heatmap and hierarchical clustering of gene expression. Pearson correlation coefficients between replicates are shown from 1 (blue) to 0.65 (white). Experiments numbered 1–3 represent biological replicates for each condition. (b,c) Volcano plots depicting significantly (adjusted p < 0.05) upregulated genes (green) and downregulated genes (red): (b) shows iPSCs (n = 3) versus dhBMECs (n = 6) and (c) shows fresh dhBMECs (n = 3) vs cryopreserved dHBMECs (n = 3). Note that raw p-values were adjusted using the Benjamini–Hochberg procedure, leading to no significant DEGs when comparing fresh versus cryopreserved.

Formation of Functional BBB Microvessels.

We have previously reported on tissue-engineered BBB microvessels that recapitulate the structure and function of brain post-capillary venules.^14,15 We constructed 150 μm diameter microvessels using fresh and cryopreserved dhBMECs by seeding cells after subculture into channels patterned within 7 mg mL⁻¹ collagen I crosslinked with genipin (Figure 5a). Fresh and cryopreserved dhBMECs reliably assembled into a confluent monolayer over 1 day after seeding into microvessel channels, forming a perfusable microvessel (Figure 5b). Microvessel permeability of Lucifer yellow was independent of the duration of dhBMEC cryopreservation (p = 0.097, non-statistically significant slope) (Figure 5c). The average 3D permeability of Lucifer yellow in non-frozen BBB microvessels was 1.76 ± 1.38 × 10⁻⁷ cm s⁻¹, while the average permeability of cryopreserved BBB microvessels was 1.15 ± 0.41 × 10⁻⁷ cm s⁻¹, a non-significant difference (p = 0.99) (Figure 5d). Additionally, Lucifer yellow permeability was much lower in cryopreserved 3D microvessels compared to 2D transwells (p < 0.001). The permeability of 10 kDa was below the detection limit for microvessels formed with both fresh and cryopreserved cells.¹⁵

Figure 5.

Cryopreserved dhBMECs form BBB microvessels with low permeability. (a) Fabrication of 3D microvessels. (b) Representative phase/fluorescence images of Lucifer yellow permeability for fresh and 1-year cryopreserved BBB microvessels. (c) Average Lucifer yellow permeability across BBB microvessels generated from fresh and cryopreserved dhBMECs. n = 4 for fresh BC1 microvessels and n = 10 for cryopreserved BC1 microvessels, across at least four independent differentiations per condition. (d) Comparison of Lucifer yellow permeability across 2D and 3D models for non-frozen and 1-year cryopreserved dhBMECs. ***p < 0.001.

DISCUSSION

Protein expression and localization of BBB and endothelial markers in confluent monolayers of dhBMECs is well conserved following 1-year cryopreservation, extending results from a previous study over a shorter cryopreservation length (1–2 weeks).⁹ In 2D confluent monolayers, TEER values declined only slightly over 1 year. While both fresh and 1-year cryopreserved cells always displayed TEER above 500 Ω cm², cryopreserved cells were more likely to display TEER below 1000 Ω cm² (30% of batches) compared to fresh cells (5% of batches). The permeability of Lucifer yellow increased 3-fold following cryopreservation for 1 year, whereas the permeabilities of rhodamine, glucose, and 10 kDa dextran were the same for cryopreserved and non-frozen cells. This suggests that measurements of transcellular transport or paracellular transport for compounds larger than 10 kDa dextran (i.e., antibodies, nanoparticles, and gene therapies) display similar permeabilities between fresh and cryopreserved cells. Importantly, cryopreserved cells maintain efflux activity of p-glycoprotein supporting use of these cells for screening BBB penetration.

The increase in Lucifer yellow permeability was associated with the media change used to introduce the solute into the transwell for measurement. We hypothesize that long periods of cryopreservation increased cell sensitivity to stress induced by a media change. This was confirmed by the strong correlation between TEER recorded after a media switch and Lucifer yellow permeability, matching previous results finding constant Lucifer yellow permeability above TEER values of 900 Ω cm.²⁶ This increased sensitivity to a media change could be mediated by minor differences in gene expression or by the confounding factor of ROCK inhibition during thawing and subculture. ROCK inhibition is utilized to increase cell survival and adhesion of dhBMECs;^9,14 although it is not present during permeability measurements (only used for initial 24 h of culture), it may alter resting cytoskeletal arrangement.⁹ To decouple this effect, future studies could utilize ROCK inhibitor on fresh dhBMECs or develop alternative approaches to introducing a solute of interest (i.e., adding concentrated Lucifer yellow as a small volume into the apical compartment with gentle agitation to ensure mixing). Additionally, it is important to note that gene expression studies were conducted on cells exposed to ROCK inhibition, thus focusing on cryopreservation-mediated changes in gene expression (not those altered by ROCK inhibition).

While absolute TEER values for confluent monolayers decreased slightly with increased cryopreservation time for confluent monolayers of dhBMECs, the fractional change in TEER normalized to the value for non-frozen cells from the same differentiation showed no decrease, highlighting the variability between differentiations. A previous study showed no change in average TEER between 1-month, 3-month, and 1-year cryopreservation compared to non-frozen cells, although there was large variation between paired replicates from the same differentiation.⁹

In contrast to primary and immortalized cell sources, dhBMECs can exhibit TEER matching theoretical and animal model values.⁵ For example, a recent triculture transwell model using immortalized cells measured TEER around 130 Ω cm².²⁸ Although the authors were able to distinguish BBB permeable and non-permeable compounds, the values of Lucifer yellow permeability were more than 3-fold higher compared to values reported here for cryopreserved dhBMECs in 2D monolayers and 120-fold higher than values for cryopreserved dhBMECs in 3D microvessels. Therefore, cryopreservation of large batches of dhBMECs is a viable alternative for widespread use in BBB research.

Differentiation from iPSCs to dhBMECs results in ~10,000 DEGs that support BBB phenotype. However, long-term cryopreservation did not result in significant changes in gene expression compared to fresh cells. We conducted gene set enrichment analysis using the top 100, 250, and 500 genes that were most differentially upregulated or downregulated (but not statistically significant) to identify potential cellular functions that could be different due to cryopreservation. Processes consistently enriched in cryopreserved cells included tumor necrosis factor-mediated signaling pathway (GO:0007155), positive regulation of cytosolic calcium ion concentration involved in phospholipase C-activating G-protein coupled signaling pathway (GO:0051482), and positive regulation of I-kappaβ kinase/NF-kappaβ signaling (GO:0043123), while processes consistently enriched in fresh cells include adenylate cyclase-activating G-protein coupled receptor signaling pathway (GO:0007189) and transmembrane receptor protein tyrosine kinase signaling pathway (GO:0007169). This suggests that cryopreserved cells may display unique responsiveness to inflammatory stimuli, which should be explored in future work.

Tissue-engineered BBB models are a valuable tool for brain research as they can mimic the BBB microenvironment (i.e., cylindrical geometry and shear stress) and provide superior spatiotemporal resolution.⁸ Lucifer yellow permeability in 3D microvessels matched previous recordings within 3D microvessels formed using other iPSC lines^13,15 and was similar to in vivo recordings within rat post-capillary venules (~1–2 × 10⁻⁷ cm s⁻¹).²⁹ However, permeability of Lucifer yellow was ~40-fold lower in 3D microvessels compared to 2D assays. The lack of a monolayer boundary in the cylindrical geometry of 3D microvessels is a likely explanation. In contrast, the boundary of the monolayer at the perimeter of a transwell insert in 2D assays has been found to decrease TEER and increase permeability.¹ In addition, the permeabilities of non-frozen and cryopreserved cells are not statistically different in 3D models, suggesting that continuous perfusion circumvents changes associated with the media changes commonly used in 2D transwell assays.

Recent work has sought to incorporate multiple cell types in 3D models;^30,31 when TEER values are low, the use of co-cultured pericytes and astrocytes can increase TEER,³⁰ however, when TEER is high this effect is less pronounced or abrogated.^5,31 Indeed, dhBMECs alone achieve TEER values ~10-fold higher than immortalized BMECs without need for complex co-culture systems.^28,30 Thus, cryopreserved dhBMECs alone may be adequate for drug delivery models, particularly 3D, where permeability measurements are buffered from the effects of cryopreservation and monolayer boundaries in transwells.

In conclusion, this study finds that dhBMECs maintain phenotype following 1-year cryopreservation in 2D monolayers and 3D microvessels. Long-term cryopreserved dhBMECs offer key advantages over the on-demand differentiation of fresh dhBMECs: (1) scalability is increased by allowing generation of large batches of biological replicates that can be used over time, (2) variability can be reduced by minimizing the need for independent differentiations, (3) following experimentation, follow-up studies can be conducted, and (4) collaborations across different laboratories can be completed from the same cell stock. Disadvantages of long-term cryopreserved dhBMECs are: (1) a loss of cell yield over time (approaching 50% at 1 year), (2) a loss of efficiency in producing monolayers with TEER above 1000 Ω cm², and (3) the unique responsiveness to media switches in 2D assays (which can alter paracellular permeability of small molecular-weight compounds). The use of 3D microvessel models can overcome these issues as they are buffered from effects of media switches and display lower and physiological permeability values. Therefore, cryopreserved dhBMECs can be a critical tool for modeling BBB phenomena, including drug delivery into the brain.

ABBREVIATIONS

BBB

blood–brain barrier

dhBMECs

iPSC-derived human BMECs

TEER

transendothelial electrical resistance

3D

three-dimensional

DEG

differentially expressed genes